The problem of smoothing data trough a transform in the Fourier domain is analyzed in the case of correlated noise affecting data. A regularization method and two GCV-type criteria are resorted in order to solve the problem, in analogy with the case of uncorrelated noise. All convergence theorems stated for uncorrelated noise are here generalized to the case of correlated noise. Numerical experiments on significant test functions are shown. © 2000 IMACS/Elsevier Science B.V. All rights reserved.

The phase-separation kinetics of binary fluids in shear flow is studied numerically in the framework of the continuum convection-diffusion equation based on a Ginzburg-Landau free energy. Simulations are carried out for different temperatures both in d=2 and 3. Our results confirm the qualitative picture put forward by the large-Rr limit equations studied by Corberi et al. [Phys. Rev. Lett. 81, 3852 (1998)]. In particular, the structure factor is characterized by the presence of four peaks whose relative oscillations give rise to a periodic modulation of the behavior of the rheological indicators and of the average domains sizes. This peculiar pattern of the structure factor corresponds to the presence of domains with two characteristic thicknesses, whose relative abundance changes with time.

Results are presented for the phase separation process of a binary mixture subject to a uniform shear flow quenched from a disordered to a homogeneous ordered phase. The kinetics of the process is described in the context of the time-dependent Ginzburg-Landau equation with an external velocity term. The large-n approximation is used to study the evolution of the model in the presence of a stationary flow and in the case of an oscillating shear. For stationary flow we show that the structure factor obeys a generalized dynamical scaling. The domains grow with different typical length scales R-x and R-perpendicular to, respectively, in the flow direction and perpendicularly to it. In the scaling regime R(perpendicular to)similar to t(alpha perpendicular to) and R(x)similar to gamma(alpha x) (with logarithmic corrections), gamma being the shear rate, with alpha(x)=5/4 and alpha(perpendicular to)=1/4. The excess viscosity Delta eta after reaching a maximum relaxes to zero as gamma(-2)t(-3/2). Delta eta and other observables exhibit logarithmic-time periodic oscillations which can be interpreted as due to a growth mechanism where stretching and breakup of domains occur cyclically. In the case of an oscillating shear a crossover phenomenon is observed: Initially the evolution is characterized by the same growth exponents as for a stationary flow. For longer times the phase-separating structure cannot align with the oscillating drift and a different regime is entered with an isotropic growth and the same exponents as in the case without shear.

The effect of shear flow on the phase-ordering dynamics of a binary mixture with field-dependent mobility is investigated. The problem is addressed in the context of the time-dependent Ginzburg-Landau equation with an external velocity term, studied in self-consistent approximation. Assuming a scaling ansatz for the structure factor, the asymptotic behavior of the observables in the scaling regime can be analytically calculated. All the observables show log-time periodic oscillations which we interpret as due to a cyclical mechanism of stretching and break-up of domains. These oscillations are damped as consequence of the vanishing of the mobility in the bulk phase.

organizzazione scientifica dell'evento

organizzazione scientifica dell' evento con revisione dei sommari

We develop a numerical analysis of the buoyancy driven natural convection of a fluid in a three dimensional shallow cavity (4.1.1) with a horizontal gradient of temperature along the larger dimension. The fluid is a liquid metal (Prandtl number equal to 0.015) while the Grashof number (Gr) varies in the range 100,000-300,000. The Navier-Stokes equations in vorticity-velocity formulation have been integrated by means of a linearized fully implicit scheme. The evaluation of fractal dimension of the attractors in the phase space has allowed the detection of the chaotic regime. The Ruelle-Takens bifurcation sequence has been observed as mechanism for the transition to chaos: the quasi periodic regime with three incommensurate frequencies is the instability mechanism responsible for the transition to chaos. Physical experiments confirm the existence of this scenario.

We present an investigation of epsilon -entropy, h(epsilon), in dynamical systems, stochastic processes and turbulence, This tool allows for a suitable characterization of dynamical behaviours arising in systems with many different scales of motion. Particular emphasis is put on a recently proposed approach to the calculation of the epsilon -entropy based on the exit-time statistics. The advantages of this method are demonstrated in examples of deterministic diffusive maps, intermittent maps, stochastic self- and multi-affine signals and experimental turbulent data. Concerning turbulence, the multifractal formalism applied to the exit-time statistics allows us to predict that h(epsilon) similar to epsilon (-3) for velocity-time measurement. This power law is independent of the presence of intermittency and has been confirmed by the experimental data analysis. Moreover, we show that the epsilon -entropy density of a three-dimensional velocity field is affected by the correlations induced by the sweeping of large scales. (C) 2000 Elsevier Science B.V. All rights reserved.

An efficient approach to the calculation of the E-entropy is proposed. The method is based on the idea of looking at the information content of a string nf data hv annalyzing the signal only nt thp instants when the fluctuations are larger than a certain threshold is an element of, i.e., by looking at the exit-time statistics. The practical and theoretical advantages of our method with respect to the usual one are shown by the examples of a deterministic map and a self-affine stochastic process.

We review a recently proposed approach to the computation of the E-entropy of a given signal based on the exit-time statistics, i.e., one codes the signal by looking at the instants when the fluctuations are larger than a given threshold, epsilon. Moreover, we show how the exit-times statistics, when applied to experimental turbulent data, is able to highlight the intermediate-dissipative-range of turbulent fluctuations. (C) 2000 Elsevier Science B.V. All rights reserved.

We consider a Cauchy singular integral equation on the real line. A direct numerical mehod for solving this integral equation is given. We prove the convergence of the proposed method.